Switching sites to perform maintenance

Nonstop Database, HiRDB Version 9 Disaster Recovery System Configuration and Operation Guide

5.3 Switching sites to perform maintenance

This section explains the procedure for switching sites to perform maintenance.

Procedure

Note

When you switch from the main site (the remote site) to the remote site (the main site) so that you can perform maintenance on the main site (the remote site), the integrity of the data on the site being maintained cannot be guaranteed after the switchover. Therefore, you cannot immediately switch back from the remote site (the main site) to the main site (the remote site). Before you can switch sites again, both of the following conditions must be satisfied:

The status of all paired logical volume groups must be set to PAIR
The value of the pd_rise_disaster_mode operand must be changed to normal (default value) and HiRDB must be restarted

The details of each step are described below.

Organization of this section: (1) Checking whether you can switch between sites; (2) Checking whether the HiRDB system at the main site is stopped; (3) Take control over paired logical volume groups; (4) Changing the value specified for the remote site's pd_rise_disaster_mode operand; (5) Starting the HiRDB system at the remote site

(1) Checking whether you can switch between sites

Check whether all of the conditions listed below are satisfied. If not, you cannot switch sites to perform maintenance.

The site switchover method in 5.1.2 Site switchover methods that can be used while the main site is running is either Switching sites to test disaster preparedness or Switching sites to perform maintenance.
The result of switching sites to perform maintenance in 5.1.4 Results of switching sites to perform maintenance is either Can be switched over without data loss or Can be switched over with data loss.
HiRDB is running with normal (default value) specified in the pd_rise_disaster_mode operand.

(2) Checking whether the HiRDB system at the main site is stopped

Check whether the HiRDB system at the main site is stopped. If not, use the pdstop command to terminate HiRDB normally. If normal termination fails, do not attempt to switch sites to perform maintenance. Refer to the error message that was output when terminating HiRDB, eliminate the cause of error, and then terminate HiRDB normally again.

If you are using the hybrid method, normal termination might not be possible. In this case, use the pdstop -f command to forcibly terminate HiRDB. For cases in which HiRDB cannot be normally terminated, see 6.2 Collecting synchronization point dumps (when using the hybrid method).

(3) Take control over paired logical volume groups

Use the horctakeover command to take control over all paired logical volume groups. If the takeover fails, refer to RAID Manager's error log to eliminate the cause of the takeover failure. Then, re-execute the takeover.

If Swap-Takeover, SVOL-Takeover, and SVOL-SSUS-Takeover were successful for all paired logical volume groups (the return value of the horctakeover command is 1, 2, or 5), execute the pairvolchk command on each paired logical volume group to check the paired logical volume status. If Swap-Takeover, SVOL-Takeover, and SVOL-SSUS-Takeover were successful for all paired logical volume groups, you can proceed to switch sites to perform maintenance.

(4) Changing the value specified for the remote site's pd_rise_disaster_mode operand

(5) Starting the HiRDB system at the remote site

Use the pdstart command to start the HiRDB system at the remote site. If the KFPS05210-I message is output and the HiRDB start processing finishes, switching sites to perform maintenance is complete.

If HiRDB cannot be started, refer to the error message that was output during HiRDB start processing, eliminate the cause of error, and then restart HiRDB.